KR19990056309A - Voice Data Storage Using Mu-Low Coding and Memory Partitioning - Google Patents

Abstract

According to the present invention, in a system for compressing and restoring voice data, the voice data may be compressed by mu-LAW coding, and the compressed voice data may be recompressed and stored through memory division by a shift function. The present invention relates to a method of storing voice data using a μ-LAW coding and memory partitioning. It is designed to compress and store four voice sampling data or two voice sampling data at one address of 32-bit or 16-bit memory by using. Thus, by using μ-LAW coding and memory partitioning, It can compress data and increase the storage efficiency of voice data. When dealing with audio signals, it has an active memory of 32-bit system, resulting in 400% memory utilization efficiency.

Description

Voice Data Storage Using Mu-Low Coding and Memory Partitioning

According to the present invention, in a system for compressing and restoring speech data, the speech data is compressed by mu-row (hereinafter, referred to as 'μ-LAW') coding, and the compressed speech data is divided through memory division by a shift function. The present invention relates to a method of storing voice data using μ-LAW coding and memory partitioning to be recompressed and stored.

In general, the real-time signal processing system requires a high-speed microprocessor and memory for fast processing, and expensive system construction is inevitable, but the memory can be reduced in size depending on the program.

In particular, in the field of dealing with voice data, since a lot of memory is allocated to the voice data, it is important to use the memory efficiently.

On the other hand, conventionally, 16-bit audio data sampled at 8 KHz by A / D conversion through a codec is stored in a 16-bit memory as it is.

However, in the conventional voice data storage method as described above, since only one voice sampling data is stored at one address of the memory, a large number of memories are required to store the voice data, and the size of the memory becomes large. In addition to the storage efficiency of the deterioration as well as the size of the entire system was also a problem.

The present invention has been made to solve the above problems, the object of which is to compress the 16-bit speech data into 8 bits by μ-LAW coding, and the four sampling data or two through the memory partition by the shift function The purpose is to provide a method of storing voice data using μ-LAW coding and memory partitioning to maximize the memory usage by compressing and storing sampling data in 32-bit or 16-bit memory.

In order to achieve the above object, the voice data storage method using μ-LAW coding and memory partitioning of the present invention compresses 16-bit voice data compressed through a codec into 8-bit voice data through a μ-LAW compressor, Next, four voice sampling data or two voice sampling data are compressed and stored in one address of 32-bit or 16-bit memory using the shift function of the microprocessor.

1 is a hardware configuration diagram for implementing a voice data storage method using mu-row coding and memory partitioning according to the present invention;

2 is a flowchart of storing voice data using mu-row coding and memory partitioning according to the present invention;

<Description of the symbols for the main parts of the drawings>

1: codec 2: compressor

3: shifter 4: memory

5: restorer

Hereinafter, a method of storing voice data using μ-LAW coding and memory partitioning according to the present invention will be described in detail with reference to the accompanying drawings.

1 is a hardware configuration diagram for implementing a voice data storage method using μ-LAW coding and memory partitioning according to the present invention, which is encoded into 16 bits after A / D conversion of a voice signal or the voice data is converted into an original voice signal A codec 1 for decoding, a compressor 2 for compressing 16-bit speech data encoded in the codec 1 into 8-bit speech data through μ-LAW coding, and a compressed in the compressor 2 Shift 8-bit voice data to store 4 voice data or 2 voice data in one address of 32-bit or 16-bit memory 4, or shift the voice data stored in memory 4 in reverse A shifter (3) and a restorer (5) for restoring the 8-bit audio data output from the shifter (3) to 16-bit audio data and outputting the same to the codec (1).

The μ-LAW coding technique used in the present invention utilizes the characteristics of the human auditory organs, and when a person hears a sound, it is algebraically (Logarithmcally).

In other words, when recording sound, loud voices do not have to have equal resolution compared to small voices.

Therefore, the voice distortion due to voice compression / restore can be minimized by placing a heavy weight on the small voice.

When the μ-LAW coding technique is formulated, Equation 1 below.

Referring to the flowchart of FIG. 2, a method of storing voice data using μ-LAW coding and memory partitioning according to the present invention on the hardware configured as described above is as follows.

First, when a voice signal is compressed and output from the codec 1 into 16-bit voice data, the compressor 2 inputs 16-bit voice data output from the codec 1 (S1) to operate the μ-LAW compressor. Through 8-bit voice data is compressed (S2).

Then, using the shift function of the shifter 3, four or two pieces of 8-bit audio data compressed by the compressor 2 are collected, that is, divided into 32 bits or 16 bits (S3), and thus a 32-bit or 16-bit memory. Store at one address (4) (S4).

For example, when the memory 4 is a 32-bit memory, the 8-bit voice data compressed by the compressor 2 is shifted four times by eight bits by using the microprocessor's shift function. The compressed voice data is stored in one address of the memory 4.

Thereafter, the voice data stored in the memory 4 may be restored by performing the above operation in reverse.

At this time, the start and end of the voice data has a separate counter variable so that only the desired voice can be output at the time of restoration.

As described above, the present invention can be used not only in 8-bit memory but also in 16-bit and 32-bit active and inactive memories.

As described above, the present invention has the effect of compressing the voice data without data loss and increasing the storage efficiency of the voice data by using the μ-LAW coding and memory partitioning, especially in real time signal processing. In case of dealing with voice signal, the 32-bit system has active memory, which has the effect of 400% memory utilization efficiency.

Claims (1)

In the system for compressing and storing voice data, Compressing the 16-bit compressed speech data into 8-bit speech data through μ-LAW coding, and splitting the compressed 8-bit speech data into 32-bit or 16-bit using the microprocessor's shift function. A method of storing speech data using mu-LAW coding and memory partitioning, comprising: storing in 32-bit or 16-bit memory.